A method for optimizing configuration of test resources of a chassis tester

Abstract

The application belongs to the technical field of automobile testing, and is used to solve the problem that the prior art can only perceive the deviation of a single target, but cannot identify the propagation path and accumulation degree of internal stress, and specifically relates to a chassis testing machine test resource optimization configuration method, which comprises: obtaining running state parameters of multiple target dimensions in the running process of the chassis testing machine; based on the deviation of the running state parameters of each target dimension from the corresponding preset reference value, the current stretching amount of each target dimension is calculated respectively; based on the stress source dimension, the current stretching amount of each target dimension, and the preset coupling stiffness between adjacent target dimensions, the stress propagation path is determined, and at least one compressible target dimension based on the stress propagation path; the application can accurately distinguish between incidental disturbance and persistent overload, and only triggers intervention when the accumulated stress exceeds the threshold value, thereby fundamentally avoiding strategy shock caused by instantaneous jitter and improving the stability of resource allocation.

Description

Technical Field

[0001] This invention belongs to the field of automotive testing technology, specifically a method for optimizing the configuration of testing resources for a chassis testing machine. Background Technology

[0002] Chassis testing machines are key equipment in automotive R&D and production verification. They are used to reproduce the various loads that vehicles experience during actual road driving under indoor conditions to verify durability, reliability, and performance. During testing, the system needs to simultaneously consider multiple objectives: ensuring the integrity and undistorted details of the acquired signals, maintaining low-latency response between control commands and data feedback, meeting the test specifications' requirements for complete storage of raw data, and keeping the instantaneous power consumption and resource usage of the control equipment within safe limits. These four objectives share the same underlying resource pool, including computing power, cache bandwidth, storage I / O, and power consumption; therefore, there are inherent conflicts and constraints between their implementation.

[0003] Existing technologies typically employ a linear approach combining independent adjustment and priority ranking when handling such multi-objective resource allocation. Specifically, the system sets an independent expected range or threshold for each objective. When the measured value of an objective deviates from its preset range, the corresponding resource parameters (such as sampling rate, data processing pipeline mode, storage strategy, etc.) are independently corrected. When multiple objectives trigger adjustment requests simultaneously, the order of resource allocation is determined according to a preset static priority. This approach can maintain basic operation under steady-state conditions, implicitly assuming that the adjustment requests of each objective are mutually isolated and linearly superimposed, and that resource conflicts can be properly resolved through priority ranking.

[0004] However, when test conditions change abruptly (such as the sudden occurrence of a high-frequency impact event) or system resources enter the critical region, the limitations of the aforementioned linear model become apparent. The fundamental problem lies in neglecting the nonlinear coupling relationship formed between the four objectives through a shared resource pool. When the resource demand of one objective increases, this demand is transmitted to adjacent objectives through resource contention, and then propagates along the conflict chain between objectives, forming a continuously accumulating "internal stress" in the system. Since existing technologies can only sense the deviation of a single objective, but cannot identify the propagation path and accumulation degree of this internal stress, when the system faces conflicting adjustment instructions issued by multiple objectives simultaneously, it will fall into a state of alternating buffer thrashing, processing latency spikes, storage congestion, and power consumption spikes. This state is not a complete collapse, but manifests as a "spasmodic paralysis" in which all objectives fail to reach the expected level. The root cause is the lack of a mechanism that can sense the intensity of dynamic conflicts between multiple objectives and actively guide the dissipation of internal stress. Summary of the Invention

[0005] The purpose of this invention is to provide an optimized configuration method for chassis testing machine testing resources, which solves the problem that existing technologies can only sense the deviation of a single target, but cannot identify the propagation path and accumulation degree of internal stress. The technical problem to be solved by this invention is: how to provide an optimized configuration method for chassis testing machine test resources that can sense the intensity of dynamic conflict between multiple targets and actively guide the dissipation of internal stress.

[0006] The objective of this invention can be achieved through the following technical solutions: A method for optimizing the configuration of test resources for a chassis testing machine includes: Acquire the operating status parameters of multiple target dimensions during the operation of the chassis testing machine; calculate the current tensile amount of each target dimension based on the deviation between the operating status parameters of each target dimension and the corresponding preset benchmark value. Based on the historical tensile sequence of each target dimension, the cumulative stress that exceeds the corresponding elastic limit in the past preset time period is calculated. When the cumulative stress of at least one target dimension is greater than the preset stress threshold, the target dimension with the largest cumulative stress is determined as the stress source dimension. Based on the stress source dimension, the current tensile amount of each target dimension, and the preset coupling stiffness between adjacent target dimensions, the stress propagation path is determined. The stress propagation path is an ordered sequence of at least one target dimension that starts from the stress source dimension, propagates along the ring topology, and passes through in sequence. Based on the stress propagation path control, a pre-set relaxation vernier slides on the annular topology, and at least one compressible target dimension is determined from multiple target dimensions according to the sliding position of the relaxation vernier. Perform step-by-step compression adjustments on the resource configuration parameters corresponding to the compressible target dimension. The adjustment range of each step-by-step compression adjustment shall not exceed a preset percentage of the current value.

[0007] The present invention has the following beneficial effects: 1. By constructing a dual-layer sensing mechanism of tensile amount and cumulative stress, this invention overcomes the limitation of existing technologies that can only identify instantaneous deviations. In existing technologies, systems often overreact to short-term fluctuations, leading to frequent oscillations in resource allocation. This invention first quantifies the real-time deviations of each target dimension as tensile amount, and further accumulates historical overload amplitudes through a circular queue to form cumulative stress that can characterize continuous pressure. As a result, the system can accurately distinguish between occasional disturbances and continuous overloads, and only triggers intervention when the cumulative stress exceeds a threshold, fundamentally avoiding strategy oscillations caused by instantaneous jitter and improving the stability of resource allocation. 2. By introducing an elastic coupling ring structure and propagation impedance calculation, this invention breaks away from the linear processing paradigm in existing technologies that treat multiple targets as independent parallel relationships and rely on static priority ordering. In existing technologies, when multiple targets simultaneously trigger adjustment requirements, the system can only process them according to a preset fixed priority order and cannot identify the stress propagation path between targets. However, this invention organizes the four target dimensions into a closed-loop topology, characterizes the conflict intensity between adjacent targets through coupling stiffness, and dynamically calculates the propagation impedance based on real-time tensile amounts, thereby accurately locating the stress source dimension and determining the main direction of stress propagation. As a result, the system no longer blindly compresses all overloaded targets, but instead directs compression resources to key dimensions on the stress propagation path, achieving a fundamental shift from static priority stacking to dynamic strain coordination. 3. By employing a relaxed cursor's directional sliding and step-by-step compression mechanism, this invention solves the problem of sudden throughput drop caused by global degradation in the resource critical zone in existing technologies. In existing technologies, when resource pressure exceeds a threshold, a one-size-fits-all strategy such as global downsampling or overall frequency reduction is often adopted, causing simultaneous damage to all targets. In contrast, this invention uses a relaxed cursor's gradual sliding on a ring-shaped topology, performing step-by-step adjustments on only a single compressible target dimension at a time, not exceeding 10% of the current value, supplemented by a compression window to limit the adjustment frequency of a single dimension. Thus, the system can release resource bottlenecks in a localized and gradual manner, avoiding the secondary impact of step adjustments on the system while ensuring that the basic performance of key dimensions is not over-compressed, maintaining a predictable level of degraded service in the resource critical zone. 4. By using differentiated recovery priority scheduling based on compression history, this invention overcomes the problem of uneven resource allocation caused by the fixed order of the recovery phase in existing technologies. In existing technologies, when the system exits an overload state, each target dimension is usually restored sequentially according to a fixed ring order, resulting in the most severely compressed dimension not receiving priority restoration. In contrast, this invention records the number of compressions and the cumulative compression magnitude of each target dimension during the compression phase, and generates a recovery priority weight based on this during the recovery phase. It then controls the relaxed cursor to slide in reverse order according to the weight, prioritizing the restoration of the dimension under the greatest pressure. As a result, the system can dynamically adjust the recovery order according to the actual resource release level of each dimension during the compression phase, achieving optimization from fixed-order restoration to on-demand weighted restoration, significantly improving the fairness of resource allocation among multiple targets and the efficiency of system state restoration. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 This is the main flowchart of Embodiment 1 of the present invention; Figure 2 This is a sub-flowchart of step compression in Embodiment 1 of the present invention. Detailed Implementation

[0010] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0011] In the actual operation of chassis testing machines, especially on complex equipment such as multi-axis hydraulic road simulation test benches, the system needs to simultaneously consider four core objectives: ensuring the fidelity and detail integrity of the acquired signals, maintaining low-latency response between control commands and data feedback, meeting the compliance requirements of test specifications for complete storage of raw data, and controlling the instantaneous power consumption and resource consumption of the equipment itself within safe thresholds. These four objectives share the same underlying resource pool, including computing power, cache bandwidth, storage input / output, and power consumption, so there are natural conflicts and constraints between their implementation.

[0012] Existing technologies typically employ a linear approach combining independent adjustment and priority ranking when handling such multi-objective resource allocation. Specifically, the system sets an independent expected range or fixed threshold for each objective. When the measured value of an objective deviates from its preset range, the resource parameters corresponding to that objective are independently corrected. When multiple objectives trigger adjustment requests simultaneously, the order of resource allocation is determined according to a preset static priority. This approach can maintain basic operation under steady-state conditions, implicitly assuming that the adjustment requests of each objective are isolated and linearly superimposed, and that resource conflicts can be properly resolved through priority ranking.

[0013] However, when test conditions change abruptly (e.g., a sudden high-frequency impact event occurs) or system resources enter the critical region, the limitations of the above linear mode become apparent. The fundamental problem lies in ignoring the nonlinear coupling relationship formed between the four targets through a shared resource pool. When the resource demand of one target increases, this demand will be transmitted to adjacent targets through resource contention, and then propagate along the conflict chain between targets, forming a continuously accumulating internal stress in the system. Since the existing technology can only sense the deviation of a single target, but cannot identify the propagation path and accumulation degree of this internal stress, when the system faces conflicting adjustment instructions issued by multiple targets at the same time, it will fall into a state of alternating cache thrashing, processing latency spikes, storage congestion, and power consumption spikes.

[0014] For example, when a high-frequency impact event occurs during endurance testing, the fidelity target requires an immediate increase in the sampling rate to capture the impact front, which leads to data queue backlog and a rapid deterioration in the latency index of the responsiveness target. After detecting the risk of buffer overflow, the compliance module forcibly starts data write to disk, further seizing storage bandwidth. The instantaneous power consumption then exceeds the safety threshold, triggering frequency reduction protection. In this process, the existing technology simultaneously issues conflicting instructions to four targets, but cannot determine which target's demand is a real overload and which is a passive transmission. Ultimately, none of the targets can reach the expected level, and the system neither crashes nor operates effectively, falling into a global low-level lock-in state.

[0015] If the above problems are not resolved, the system will continue to lose its ability to objectively distinguish dynamic conflicts between multiple targets. In particular, the failure to identify passively transmitted overloads will cause the system to apply ineffective compression to the dimension where the real pressure source is located, causing resources that should be released first to be delayed. At the same time, the disorderly spread of stress throughout the entire loop will cause the resource allocation strategy to oscillate at high frequency among multiple targets, making the system throughput lower than the stable degraded state. As a result, the inaccuracy of resource allocation will systematically weaken the operational stability of the test machine under critical conditions, affecting the data integrity and control reliability of complex test tasks under extreme conditions.

[0016] Example 1: As Figure 1-2 As shown, a method for optimizing the configuration of test resources for a chassis testing machine includes: Step S1: Obtain the operating status parameters of multiple target dimensions during the operation of the chassis testing machine; calculate the current tensile amount of each target dimension based on the deviation between the operating status parameters of each target dimension and the corresponding preset benchmark value. In step S1, this embodiment first acquires real-time operating status parameters of multiple target dimensions of the chassis testing machine during operation at a fixed sampling period. The sampling period is synchronized with the real-time control period of the testing machine; in this embodiment, it is preset to 10 milliseconds to ensure that the monitoring frequency of the system status is sufficient to capture dynamic changes. To achieve a comprehensive assessment of the system's multi-dimensional stress, this embodiment focuses on at least the following target dimensions: fidelity, responsiveness, compliance, and consumption rate. The operating status parameter for the fidelity dimension is the actual sampling rate read from the analog-to-digital converter configuration register, denoted as... The unit is Hertz, and the corresponding preset baseline value is the ideal sampling rate set by the current test task configuration file. The operational status parameter for the responsiveness dimension is the actual latency calculated by the difference between the packet sending timestamp and the receiving timestamp, denoted as . The unit is milliseconds, and the corresponding preset baseline value is the target delay: the maximum allowable delay of the control loop. The compliance dimension's runtime status parameter is the disk write success rate obtained through file system write return values ​​statistics, denoted as... Its value is between 0 and 1, and the corresponding preset baseline value is 100% (i.e. The operating state parameter in the power consumption rate dimension is the normalized instantaneous power consumption value obtained by filtering the current sample value from the power management chip through a moving average, denoted as . The corresponding preset baseline value is the power consumption threshold. This threshold is usually set at 80% of the device's rated power consumption to allow for a safety margin.

[0017] To suppress the impact of transient noise on subsequent calculations, after obtaining the original operating state parameters, this embodiment further performs exponentially weighted moving average filtering on three high-dynamic parameters: actual sampling rate, actual latency, and instantaneous power consumption. Specifically, for the actual sampling rate, a filtering coefficient of 0.7 is used, meaning the filtered value of the current period is a weighted composite of 70% of the current original value and 30% of the filtered value of the previous period; for the actual latency, a filtering coefficient of 0.8 is used; and for the instantaneous power consumption, a filtering coefficient of 0.9 is used. Since the disk placement success rate is based on the cumulative value over a 1-second statistical period, its changes are relatively gradual, so no inter-frame smoothing is performed. The statistical result of the most recent second is used directly, and it remains constant in each 10-millisecond period until the next statistical period update. After this preprocessing, stable parameters for each dimension used in subsequent stretching calculations are obtained, denoted as follows: , and .

[0018] After obtaining the aforementioned stable parameters, this embodiment calculates the current stretch amount for each target dimension based on the deviation between the operating state parameters of each target dimension and the corresponding preset benchmark value. This stretch amount is defined as the relative degree to which the dimension deviates from the ideal state, and its value is limited to a closed interval of 0 to 1, where 0 indicates complete satisfaction of the benchmark and 1 indicates extreme deviation. The specific calculation method for each stretch amount is as follows.

[0019] Regarding the fidelity dimension, fidelity stretching Calculated using the following formula:

[0020] ; The physical meaning of this formula lies in quantifying the degree of sampling accuracy loss: when the actual sampling rate is equal to or higher than the ideal sampling rate, the stretching is 0; when the actual sampling rate drops to 0, the stretching is 1; in intermediate cases, there is a linear relationship. For example, suppose the ideal sampling rate is 10kHz, and the actual sampling rate after filtering is 8kHz, then... This indicates that there is a 20% resource deficit in this dimension.

[0021] For the responsiveness dimension, responsive stretching Calculated using the following formula:

[0022] ; This formula reflects the degree of degradation in the system's current latency response: the stretch is 0 when the actual latency does not exceed the target latency; the stretch is 1 when the actual latency reaches twice the target latency. For example, if the target latency is 5ms and the actual latency after filtering is 8ms, then... .

[0023] For the compliance dimension, compliance stretch amount Calculated using the following formula:

[0024] ; in This represents the disk write success rate in the most recent second. When the success rate is 100%, the stretch is 0; when all data packets fail, the stretch is 1. For example, if 1000 data packets should have been written to disk in the most recent second, and 950 were successfully written, then... , .

[0025] Regarding the consumption rate dimension, the consumption rate stretching amount Calculated using the following formula:

[0026] ; This formula characterizes how close the current power consumption is to the danger zone: when the instantaneous power consumption is below the threshold, the stretch is 0; when the instantaneous power consumption reaches the rated value (i.e., the normalized value is 1), the stretch is 1. For example, assuming the rated power consumption normalized value is 1, the power consumption threshold is 0.8, and the current instantaneous power consumption after filtering is 0.9, then... .

[0027] Based on the above calculations, this embodiment quantifies the real-time status of the four dimensions into fidelity stretch amounts. Response elongation , Follow the degree of stretching and consumption rate stretch amount These four dimensionless values ​​together constitute the stretch vector of the current sampling period. This serves as the foundational input for subsequent cumulative stress assessment and stress propagation analysis. It is noteworthy that the calculation of each tensile quantity is based entirely on directly measurable physical quantities, without relying on any subjective judgments or empirical parameters, ensuring the objectivity and reproducibility of the system state assessment. Furthermore, all calculation results are strictly truncated to the [0,1] interval, avoiding divergence in subsequent calculations due to extreme outliers, thus laying a data foundation for the stable operation of the entire resource allocation optimization method.

[0028] Step S2: Based on the historical tensile sequence of each target dimension, calculate the cumulative stress that represents the total amplitude of the corresponding elastic limit in each target dimension over the past preset time period; when the cumulative stress of at least one target dimension is greater than the preset stress threshold, the target dimension with the largest cumulative stress is determined as the stress source dimension. In step S2, this embodiment calculates the cumulative stress characterizing the degree of continuous overload of each target dimension during the historical period based on the stretching amount of each target dimension obtained in step S1 during the current sampling period and the historical stretching amount sequence of each target dimension over a preset time period. The core of this step is to transform the instantaneous deviation into a cumulative index reflecting long-term pressure, so as to avoid overreacting to short-term fluctuations.

[0029] Specifically, this embodiment first presets an elastic limit for each target dimension. This elastic limit is a fixed threshold between 0 and 1, used to define the tolerable normal stretching range of that dimension. In this embodiment, based on engineering experience from typical chassis testing machines, the elastic limits for each dimension are set as follows: Elastic limit of the fidelity dimension. This means that the sampling rate can be reduced by up to 30% without triggering overload accumulation; the elastic limit of the responsiveness dimension. This means that the allowed latency is up to 20% higher than the target value; adhering to the elasticity limit in the degree dimension. This means that the failure rate for disk placement is allowed to not exceed 25%; the elastic limit in terms of consumption rate. This means that power consumption is allowed to exceed the safety threshold by 15%. The above flexible limit value can be configured and adjusted according to different test tasks or equipment characteristics, but once set, it remains fixed during system operation.

[0030] After obtaining the current stretching amount for each dimension, this embodiment further maintains a fixed-length circular queue for each target dimension to store the instantaneous overload amount for each sampling period within a preset time period. In this embodiment, the preset time period is set to 10 seconds, and the sampling period is 10 milliseconds; therefore, the length of the circular queue is... Each queue element stores the instantaneous overload of that dimension during the corresponding sampling period. This instantaneous overload is defined as the positive value of the current stretch of that dimension minus its elastic limit; if the difference is less than or equal to 0, it is set to 0. ; in Indicates the target dimension. This indicates the current sampling period number. The physical meaning of this definition is that only the portion of the stretching exceeding the elastic limit is considered "overload" and included in the cumulative stress, while normal stretching fluctuations are ignored, thus filtering out the small fluctuations of the system under normal operating conditions.

[0031] When the system starts, all circular queues are initialized to all zeros, and the initial accumulated stress is 0. In each sampling period, this embodiment performs the following operations: First, the instantaneous overload of the current sampling period is... Push the value to the tail of the circular queue corresponding to the given dimension, and simultaneously pop the value from the head of the queue that was 10 seconds ago. This ensures that the queue always stores the instantaneous overload sequence of the most recent 10 seconds. Then, the sum of all elements in the queue is calculated as the cumulative stress of that dimension at the current moment. ,Right now:

[0032] ; in This is the index of the element in the queue. The physical meaning of this accumulated stress lies in quantifying the total magnitude of the excess over the elastic limit accumulated in this dimension over the past 10 seconds, and its unit is the dimensionless "overload·second" equivalent integral. For example, if the stretch in the fidelity dimension is 0.4 for 5 consecutive seconds, and the elastic limit is 0.3, then the instantaneous overload every 10 milliseconds is 0.1, and the cumulative contribution over 5 seconds is... This value is a part of the cumulative stress.

[0033] After completing the cumulative stress calculation in four dimensions, this embodiment obtains the cumulative stress vector. To determine whether the system is currently in a stress state requiring intervention, this embodiment presets a stress threshold. This threshold value is extremely small and is used to characterize a relaxed state where the accumulated stress is almost negligible. When the accumulated stress in all target dimensions is less than this stress threshold, i.e. Simultaneously, in this embodiment, the system is marked as "relaxed state," indicating that the overload in all dimensions has not continued to accumulate to the point where intervention is required. At this time, all subsequent adjustment steps are skipped, the system maintains the existing resource configuration unchanged, and waits for the next sampling period to re-evaluate.

[0034] Conversely, if the cumulative stress in at least one target dimension is greater than or equal to the stress threshold, this embodiment marks the system as "stressed," indicating that the system has experienced persistent resource pressure over a period of time. In this state, this embodiment further identifies the target dimension with the highest cumulative stress from the four dimensions and determines it as the stress source dimension. If the cumulative stress in multiple dimensions is equal and all are positive, the dimension with the highest priority is selected as the stress source according to the order on the ring topology. In this embodiment, the priority order is set as follows: fidelity dimension takes precedence over responsiveness dimension, responsiveness dimension takes precedence over compliance dimension, and compliance dimension takes precedence over consumption rate dimension. The location of the stress source dimension provides a clear starting point for subsequent analysis of the stress propagation path.

[0035] Through the above processing, this embodiment transforms the instantaneous tensile amount obtained in step S1 into cumulative stress that can reflect the continuous overload pressure, and thereby realizes the quantitative discrimination between the system state and the stress state, providing a reliable data basis for step S3 and subsequent resource compression decisions.

[0036] Step S3: Based on the stress source dimension, the current tensile amount of each target dimension, and the preset coupling stiffness between adjacent target dimensions, determine the stress propagation path. The stress propagation path is an ordered sequence of at least one target dimension that starts from the stress source dimension, propagates along the ring topology, and passes through in sequence. In step S3, this embodiment determines the stress propagation path based on the stress source dimension determined in step S2, the current tensile amount of each target dimension calculated in step S1, and the pre-calibrated coupling stiffness between adjacent target dimensions. This path is defined as an ordered sequence of at least one target dimension that starts from the stress source dimension, propagates along the ring-shaped topology, and passes through sequentially, and is used to guide the allocation direction of subsequent compression resources.

[0037] Before performing this step, this embodiment first completes the offline calibration of coupling stiffness and the preset of the ring topology. The order of adjacent target dimensions on the ring topology is as follows: fidelity dimension, responsiveness dimension, compliance dimension, and consumption rate dimension, with the consumption rate dimension adjacent to the fidelity dimension, forming a closed loop. On this ring, each target dimension has an adjacent pair with its clockwise and counterclockwise adjacent dimensions, i.e., fidelity is adjacent to responsiveness, responsiveness is adjacent to compliance, compliance is adjacent to consumption rate, and consumption rate is adjacent to fidelity.

[0038] For the four pairs of adjacent target dimensions mentioned above, this embodiment obtains their respective coupling stiffness through offline calibration. The calibration process is performed before system deployment, selecting no more than 20 typical test conditions, including but not limited to flat road constant speed conditions, bumpy road conditions, emergency braking conditions, and steering excitation conditions. For each pair of adjacent target dimensions, such as fidelity and responsiveness, the calibration method is as follows: the test machine is run under a certain typical condition, and the resource allocation for the fidelity dimension is increased separately (e.g., the sampling rate is increased by 10%), and the percentage change in tensile strength of the responsiveness dimension is measured; the coupling stiffness is calculated. The ratio of this change, i.e. ,in To maintain fidelity, the rate of change of stretch amount, This represents the rate of change of the responsive tensile force. The final coupled stiffness value is obtained by averaging the absolute values ​​of the calculation results under multiple operating conditions. Similarly, calibration is performed separately. , and After calibration, these coupling stiffness values ​​are stored in the system as a read-only table, and retrieved at runtime by indexing the table based on the current operating conditions.

[0039] After obtaining the stress source dimension, the current tensile amount in each dimension, and the coupling stiffness, this embodiment begins to determine the stress propagation path. First, for the stress source dimension, the first impedance value propagating in the clockwise direction and the second impedance value propagating in the counterclockwise direction are calculated. Let the stress source dimension be... The dimensions of the adjacent targets in the clockwise direction are denoted as The dimensions of adjacent targets in the counterclockwise direction are denoted as The first impedance value in the clockwise direction. Calculated using the following formula: ; Second impedance value in the counterclockwise direction Calculated using the following formula: ; in and These are the coupling stiffness between the stress source dimension and two adjacent dimensions, respectively. and These represent the current stretching amounts in two adjacent dimensions. The physical meaning of this impedance calculation formula is as follows: the greater the coupling stiffness, the easier it is for stress to propagate in that direction, hence the impedance value is positively correlated with the coupling stiffness; while the greater the current stretching amount in an adjacent dimension, the more stressed that dimension is, the more sensitive it is to the incoming stress, hence the impedance value is negatively correlated with the adjacent stretching amount, that is, the impedance value is positively correlated with the current stretching amount in an adjacent dimension. The greater the stretching amount, the greater the impedance value, indicating that the direction of the current stretching is more resistant to the incoming stress, and the stress should preferentially propagate to the direction with the smaller impedance value (i.e., the relaxed region with the smaller stretching amount).

[0040] Let's take specific numerical examples to illustrate. Assume that the stress source dimension determined in step S2 is the fidelity dimension. The adjacent dimensions in the clockwise direction are the responsiveness dimensions. The adjacent dimensions in the counter-clockwise direction are the consumption rate dimensions. .set up , Current responsive stretch amount Consumption rate stretching amount Then the first impedance value clockwise. The second impedance value in counterclockwise direction Since the counterclockwise impedance value is smaller, it means that stress is more likely to propagate in the direction of the wear rate.

[0041] After calculating the impedance values ​​in both directions, this embodiment determines the direction with the smaller impedance value as the main propagation direction. If the impedance values ​​in both directions are equal, the clockwise direction is selected as the main propagation direction by default.

[0042] After determining the main propagation direction, this embodiment starts from the stress source dimension and sequentially checks adjacent target dimensions along the main propagation direction to construct a stress propagation path. Specifically, the stress source dimension is used as the first element of the path, and then it moves along the main propagation direction to its adjacent dimensions. If the current tensile amount of the adjacent dimension is greater than or equal to its elastic limit (i.e., the dimension is in an overloaded state), the dimension is added to the propagation path, and the next adjacent dimension is checked along the same direction. If the current tensile amount of the adjacent dimension is less than its elastic limit (i.e., the dimension is not overloaded), the path construction stops and does not continue to extend. This process is repeated until the first target dimension with a current tensile amount less than its corresponding elastic limit is encountered. If no unoverloaded dimension is encountered after traversing the entire ring topology along the main propagation direction, the propagation path includes all target dimensions on the ring.

[0043] Continuing with the example above, the main propagation direction is counterclockwise, that is, from the perspective of fidelity. Towards the consumption rate dimension Propagation. Assume an elastic limit in the consumption rate dimension. Current stretch amount ,because Since the consumption rate dimension is not overloaded, the propagation path only includes the stress source dimension itself, i.e., the path is... If we assume Greater than If so, the consumption rate dimension is added to the path, and then the next dimension is checked in a counter-clockwise direction. Since the counter-clockwise adjacent dimensions of the consumption rate dimension on the ring are compliance dimensions... Let its current stretch amount Elastic limit ,because Then, the path will be added according to the degree dimension, and the path will be... Continue checking if the dimensions adjacent to the compliance dimension in the counter-clockwise direction are responsive dimensions. ,set up If it stops, the final propagation path is: .

[0044] Through the above calculations, this embodiment obtains a stress propagation path represented by an ordered sequence. This path clearly defines the overload target dimensions that originate from the stress source and pass sequentially along the main propagation direction. The end of this path is the boundary of the current stress propagation. In subsequent steps, the relaxation cursor will slide along this path direction, guiding the compression resources to preferentially act on the target dimensions on the path, thereby achieving directional dissipation of internal stress.

[0045] Step S4: Based on the stress propagation path control, a pre-set relaxation vernier slides on the annular topology, and at least one compressible target dimension is determined from multiple target dimensions according to the sliding position of the relaxation vernier; In step S4, this embodiment controls a pre-set relaxation cursor to slide on the annular topology based on the stress propagation path and its main propagation direction determined in step S3, and determines at least one compressible target dimension from multiple target dimensions according to the sliding position of the relaxation cursor. The core of this step is to transform the abstract stress propagation path into a specific, executable compression command issuance mechanism.

[0046] Before performing this step, this embodiment predefines the data structure and initial state of the relaxation cursor. The relaxation cursor is a virtual pointer pointing to the connection points between adjacent target dimensions on a ring topology. There are four connection points on the ring topology: the connection point between the fidelity dimension and the responsiveness dimension is denoted as... The connection point between the responsiveness dimension and the compliance dimension is denoted as The connection point located between the compliance dimension and the consumption rate dimension is denoted as The connection point between the consumption rate dimension and the fidelity dimension is denoted as The initial position of the relaxation cursor is set to... In addition, the system maintains a current movement direction variable for the relaxed cursor. A value of +1 indicates clockwise direction, and a value of -1 indicates counterclockwise direction; the initial value is +1. Simultaneously, the system maintains a reverse continuous count variable. The initial value is 0, which is used to record the number of times the main propagation direction is continuously opposite to the current movement direction.

[0047] Within each sampling period (10 milliseconds), if step S2 determines that the system is in a "stressed state," this embodiment executes the following sliding determination logic. First, the main propagation direction determined in step S3 is obtained. The direction takes a value of +1 to represent clockwise and a value of -1 to represent counterclockwise. Then... With the current direction of movement Compare them.

[0048] like and If they are the same, it means that the current stress propagation direction is consistent with the vernier movement direction. In this case, the counting will continue in reverse. Clear the cursor and control it to slide one step along the current movement direction, moving to the next adjacent connection point. For example, if the current cursor is located at... If the current movement direction is clockwise, the cursor will move to [location] after sliding. If the current movement direction is counterclockwise, the cursor will move to [location] after sliding. .

[0049] like and Conversely, this indicates that the current stress propagation direction is inconsistent with the vernier movement direction, and in this case, the counting will proceed in reverse. Increase by 1. When When the preset threshold number of moves is reached, this embodiment reverses the current movement direction, that is... Multiply by -1, then clear the reverse continuous count to zero, and control the relaxation cursor to slide one step along the new inverted direction. If If the preset threshold number of attempts has not been reached, the cursor will not slide and no compression marking operation will be performed in this cycle. In this embodiment, the preset threshold number of attempts is set to 5, which means that the direction is allowed to be reversed only when the main propagation direction is opposite to the current movement direction for 50 consecutive milliseconds (5 sampling cycles). The purpose of this directional inertial hysteresis mechanism is to prevent the cursor from frequently changing direction due to instantaneous jitter in the main propagation direction, thereby avoiding oscillations in the compression strategy.

[0050] In this embodiment, a compression marking operation is performed each time the relaxed cursor slides past a connection point. Specifically, for each connection point traversed by the cursor, the two target dimensions connected by that connection point are determined, and the target dimension located downstream of the main propagation direction is identified. The rule for determining the downstream direction is: when moving along the main propagation direction, the target dimension pointed to from the connection point is the downstream target dimension. For example, assuming the main propagation direction is clockwise, the cursor moves from... Slide to Passing through the connection point The connection point has a high fidelity dimension. and responsiveness dimension The downstream target dimension pointing clockwise from this connection point is the responsive dimension. Therefore, the responsiveness dimension is marked as a compressible target dimension.

[0051] Once a dimension is marked as a compressible target dimension, this embodiment simultaneously starts a compression window timer for that dimension. The duration of the compression window is a preset fixed value, set to 50 milliseconds in this embodiment. The system maintains an independent compression window timer for each target dimension, recording the validity period of the compressibility mark for that dimension. Within the validity period of the compression window, the same target dimension can undergo at most one step compression adjustment, a constraint guaranteed by the execution logic of the subsequent step S5. After the compression window expires, the system automatically clears the compressibility mark for that dimension. If compression is required again later, it must wait for the cursor to pass over again and re-mark it.

[0052] Let's illustrate the complete execution of the above process with a concrete example. Assume the system is currently in a stress state, and the current relaxation cursor is located at... The current movement direction is clockwise, and the reverse continuous count is 0. The main propagation direction determined in step S3 is clockwise. Because... and The same applies; the reverse continuous count is reset to zero, and the cursor slides clockwise to... Passing through the connection point , downstream target dimension responsiveness dimension Marked as compressible, a 50-millisecond compression window is initiated. In the next sampling period, assuming the main propagation direction changes to counterclockwise while the current movement direction remains clockwise (because the flip condition has not yet been met), the two are reversed, the reverse continuous count increases to 1, the cursor does not slide, and no new compression mark is generated. If the main propagation direction remains counterclockwise for the next 4 consecutive sampling periods, the reverse continuous count increases sequentially to 2, 3, 4, and 5. When it reaches 5, the current movement direction is reversed to counterclockwise, the reverse continuous count is reset to zero, the cursor slides in the counterclockwise direction, and a new compression mark is generated. Through the above mechanism, this embodiment realizes the directional guidance of the stress propagation path for compression resources, while ensuring the stability of the direction decision.

[0053] Step S5: Perform step compression adjustment on the resource configuration parameters corresponding to the compressible target dimension. The adjustment range of each step compression adjustment shall not exceed a preset percentage of the current value.

[0054] In step S5, this embodiment performs step-by-step compression adjustment on the resource configuration parameters corresponding to the compressible target dimension based on the compressible target dimension and its compression window state determined in step S4. The core of this step is to release the resource occupation of the target dimension in a gradual and controllable manner, thereby alleviating the current accumulated stress without excessively impacting the system stability.

[0055] Before performing compression adjustments, this embodiment first maintains an independent "last compression timestamp" variable for each target dimension to record the time when the dimension last underwent compression adjustments. When step S4 marks a target dimension as compressible and starts a compression window, the duration of the compression window is 50 milliseconds. The system checks the last compression timestamp of the dimension: if the interval between the current time and the last compression timestamp is less than 50 milliseconds, the current compression adjustment is skipped, ensuring that the same target dimension undergoes at most one step compression adjustment within 50 milliseconds; if the interval reaches or exceeds the compression window duration, or if the dimension has not yet undergone compression within this window, compression is allowed. This constraint mechanism prevents over-compression and avoids system oscillations caused by frequent adjustments.

[0056] For compressible target dimensions that are allowed to undergo compression adjustments, this embodiment performs corresponding step-by-step compression operations based on the type of the dimension. All compression operations follow the principle that the adjustment range in a single operation does not exceed 10% of the current value, thereby releasing resources in a gradual manner.

[0057] When the compressible target dimension is the fidelity dimension, this embodiment executes a two-level priority compression strategy. First, it prioritizes reducing the sampling precision (i.e., the effective number of bits in the analog-to-digital conversion), because reducing the sampling precision has a smaller impact on system latency and can more directly free up cache bandwidth and computing resources. Let the current sampling precision be... The unit is bits, and the preset minimum allowable precision is... (For example, 12 bits). The target precision for this compression adjustment is... Determined in the following ways: ; in This indicates rounding down to the nearest integer, ensuring the adjusted precision is integer. If the calculated... Greater than or equal to Then the sampling precision will be adjusted to This compression is complete; the sampling rate will not be adjusted. If the current sampling accuracy has already decreased... If this is the case, then the sampling rate will be further reduced. Let the current sampling rate be... The unit is Hertz. This represents the current actual sampling rate, i.e., the unfiltered raw value in step S1. Because the compression adjustment directly affects the hardware registers, the preset minimum allowable sampling rate is... (For example, 1kHz), then the target sampling rate for this compression adjustment is... Determined in the following ways: When the sampling accuracy or sampling rate has dropped to a preset minimum value, and the compressible target dimension is still marked, the system will no longer perform further compression to avoid the test task being unable to continue. This state will remain until the stress dissipates naturally or the test task is manually intervened.

[0058] The adjusted sampling rate and sampling precision are implemented by writing to the analog-to-digital converter's configuration register. The register switching time is less than 1 millisecond, which does not affect data continuity. For example, assuming the current sampling rate is 10kHz and the sampling precision is 16 bits, The bit is calculated during the first compression. If the sampling precision is reduced to 14 bits, and the current precision is already 12 bits, then a sampling rate adjustment will be triggered. .

[0059] When the compressible target dimension is a responsive dimension, this embodiment switches the data pipeline submission mode from "strictly ordered submission" to "unordered submission with timestamp reordering." In strictly ordered submission mode, data packets must be submitted to the output queue in strict order of acquisition time. If a previous data packet is blocked while waiting for processing, all subsequent data packets are blocked, resulting in head-of-queue blocking. After switching to unordered submission mode, the system allows subsequent data packets in the output queue to be submitted before the previous data packet has finished processing. Simultaneously, a timestamp field is retained for each data packet, allowing the receiving end to reorder them based on the timestamp after receipt. This mode switching is achieved by modifying a flag bit in the data distribution module. To achieve this, the flag bit is set from... Set as After that, the DMA descriptor's commit order constraint is removed.

[0060] When the compressible target dimension is compliance, this embodiment switches the data persistence mode from "full storage" to "keyframe plus differential storage". In full storage mode, each data packet is completely written to the disk, consuming a large amount of storage bandwidth. After switching, the system uses 100 data packets as a storage group, storing the first data packet in the group as a keyframe; for the subsequent 99 data packets, only the difference between the first and second data packets (represented by a 16-bit signed integer) is stored, and the difference sequence is compressed using run-length encoding. This mode can reduce storage bandwidth consumption by approximately 70%. The mode switch is implemented by calling the interface function of the storage driver layer, and at the same time, the internal state of the differential encoder is reset to ensure that the new mode is used for storage starting from the next data packet.

[0061] When the compressible target dimension is the consumption rate dimension, this embodiment reduces the clock frequency of non-critical computing tasks, while keeping the clock frequency of critical tasks (real-time control, data acquisition) unchanged. Let the current clock frequency of non-critical tasks be... The unit is Hertz, and the preset minimum allowable frequency is (For example, 50% of the rated frequency), then the target frequency for this compression adjustment is... Determined in the following ways:

[0062] ; Adjustments are made by writing to the CPU's P-State register or clock divider. For example, if the current non-critical task clock frequency is 200MHz, the frequency will be reduced to 180MHz after the initial compression. This operation directly affects the power consumption of background computing tasks such as spectrum analysis and peak statistics, while the response speed of the real-time control loop remains unaffected.

[0063] After each compression adjustment is completed, this embodiment updates the previous compression timestamp of the target dimension to the current time and records the magnitude of this compression (i.e., the relative rate of change of parameters before and after the adjustment, for example, the magnitude of reducing from 16 bits to 14 bits). That is, 12.5%, but because the compression step size is limited to 10%, it is actually achieved through multiple steps during execution. This record is used to determine the recovery priority in the subsequent recovery phase.

[0064] Through the aforementioned step-by-step compression adjustment, this embodiment achieves gradual resource release for compressible target dimensions. Each compression operation is performed in steps not exceeding 10% of the current value, avoiding secondary impacts on the system caused by abrupt changes in resource requirements. Simultaneously, the combination of the compression window mechanism and the previous compression timestamp ensures that the same dimension is not over-compressed due to consecutive triggers within a short period, providing a guarantee for the stable operation of the system.

[0065] After completing the aforementioned step-by-step compression adjustment, this embodiment further records compression history information for each target dimension that underwent compression adjustment, in order to achieve differentiated recovery priority scheduling in the subsequent recovery phase. Specifically, the system maintains two cumulative variables for each target dimension: a compression count. and cumulative compression Among them, the number of compressions count This is an incrementing value of 1 each time a step compression adjustment is performed on this dimension, with an initial value of 0; the cumulative compression magnitude. This is the sum of the magnitude values ​​from each compression adjustment, initially set to 0. Each time a compression adjustment is performed, the compression magnitude is recorded. This magnitude is defined as the rate of change of resource configuration parameters before and after compression. For example, when the sampling precision is reduced from 16 bits to 14 bits, the compression magnitude is... That is, 12.5%. Then update the compression count for that dimension. Cumulative compression The above records are completed within the same sampling period during the compression adjustment and are continuously stored in the system memory until the system exits the stress state and is reset.

[0066] After the compression phase is completed, when the system detects that the cumulative stress of all target dimensions is less than the preset stress threshold and the duration exceeds the preset stabilization time (5 seconds in this embodiment), this embodiment first performs the initial operation of the recovery phase: control the relaxation cursor to slide in the opposite direction of the current movement direction at a slower speed than the compression phase. Each time it slides past a connection point in the opposite direction, the compressible mark of the target dimension downstream of the connection point is released, and a step-by-step recovery adjustment is performed on the target dimension. The recovery adjustment is the inverse operation of the aforementioned step-by-step compression adjustment, and the single recovery amplitude does not exceed 10% of the current value.

[0067] Building upon the aforementioned basic recovery mechanism, this embodiment further introduces recovery priority scheduling based on compression history to optimize the processing order of each connection point during the reverse sliding process. Specifically, after entering the recovery phase, the system first counts the number of compressions for each target dimension. and cumulative compression Generate recovery priority weights for each target dimension. The recovery priority weight is positively correlated with both the number of compression attempts and the cumulative compression magnitude. In this embodiment, a weighted summation method is used for calculation.

[0068] ; in The preset compression amplification factor, set to 0.05 in this embodiment, is used to balance the contribution of a single large compression and multiple small compressions to the recovery priority. The physical meaning of this formula is: the larger the cumulative compression magnitude, or the more compression cycles, the greater the resource release pressure that dimension experiences during the compression phase, and therefore it should receive a higher recovery priority during the recovery phase.

[0069] After obtaining the recovery priority weights for each target dimension, this embodiment further determines the order in which the relaxation cursor passes through each connection point during the reverse sliding process. Specifically, the system will determine the order in which the four connection points on the ring topology pass through each connection point. , , , They are respectively associated with their downstream target dimensions, among which The downstream target dimension is the responsiveness dimension. , The downstream target dimension is the compliance dimension. , The downstream target dimension is the consumption rate dimension. , The downstream target dimension is the fidelity dimension. Then, the system sorts each connection point in descending order according to the recovery priority weight of its downstream target dimension, resulting in a reverse-sliding priority queue.

[0070] During the reverse sliding phase, this embodiment no longer slides through each connection point in a fixed order on the ring. Instead, it controls the relaxation cursor to slide backward through each connection point in the aforementioned priority queue. Each time a connection point is reverse-slid through, the system performs a step-by-step recovery adjustment on the downstream target dimension of that connection point and removes its compressibility mark. After all connection points have been reverse-slid processed, the relaxation cursor finally returns to its initial position, and the resource configuration parameters of all target dimensions are restored to their pre-compression state, at which point the system exits the stress state.

[0071] Let's illustrate the execution process of the above expansion mechanism with a concrete example. Assume that during the compression phase, the fidelity dimension is compressed 3 times, with a cumulative compression margin of 0.27; the responsiveness dimension is compressed once, with a cumulative compression margin of 0.10; and the compliance and consumption rate dimensions are not compressed. Then, the recovery priority weights for each dimension are calculated as follows: Fidelity Dimension responsiveness dimension Follow the degree dimension Consumption rate dimension Based on this, the priority order of each connection point is determined: (Downstream is the fidelity dimension, weighted at 0.42) Highest priority, followed by... (The downstream dimension is responsive, with a weight of 0.15), the other two connection points have a weight of 0 and the same priority. The system generates a priority queue. The mechanism controls the relaxed cursor to slide backwards to the aforementioned connection points, prioritizing the restoration of the fidelity dimension, followed by the responsiveness dimension, and finally processing the uncompressed dimensions. This ensures that the most compressed target dimension receives priority restoration, achieving fairness in resource allocation during the restoration phase.

[0072] Through the above-mentioned extended mechanism, this embodiment further introduces differentiated recovery priority scheduling based on compression history on top of the basic recovery steps. This enables the system to dynamically adjust the recovery order according to the pressure level of each target dimension during the compression stage when exiting the stress state, thus avoiding the problem of uneven resource allocation that may be caused by fixed-order recovery.

[0073] In summary, this embodiment, by constructing an elastically coupled ring structure anchored by fidelity, responsiveness, compliance, and consumption rate, breaks the linear processing paradigm of existing technologies that treats multiple objectives as independent parallel relationships and relies on static priority ranking. Based on this, this embodiment uses tensile amount to characterize the real-time deviation of each dimension, uses accumulated stress to filter short-term fluctuations to identify sustained overload, and then transforms the stress propagation path into directional sliding of a relaxation cursor through propagation impedance calculation and directional inertial hysteresis mechanism, achieving gradual and localized resource release for compressible target dimensions. Furthermore, during the recovery phase, differentiated recovery priority scheduling is introduced based on compression history records. Thus, this embodiment transforms the originally conflicting multi-objective resource allocation problem into an active perception and guided dissipation process of internal stress propagation paths on the ring structure. This enables the system to automatically identify pressure sources, directionally release bottleneck resources, and avoid strategy oscillations in the resource critical zone. Ultimately, it achieves a fundamental shift from static priority stacking to dynamic strain coordination, from passive response to active guidance, and from global degradation to local gradual adjustment, significantly improving the operational stability and resource utilization efficiency of the chassis test machine under multi-objective strong coupling conditions.

[0074] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

[0075] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0076] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for optimizing the allocation of test resources for a chassis testing machine, characterized in that, include: Acquire operational status parameters for multiple target dimensions during the operation of the chassis testing machine; Based on the deviation between the operating status parameters of each target dimension and the corresponding preset benchmark value, the current stretching amount of each target dimension is calculated respectively. Based on the historical tensile sequence of each target dimension, the cumulative stress that exceeds the corresponding elastic limit in the past preset time period is calculated. When the cumulative stress of at least one target dimension is greater than the preset stress threshold, the target dimension with the largest cumulative stress is determined as the stress source dimension. Based on the stress source dimension, the current tensile amount of each target dimension, and the preset coupling stiffness between adjacent target dimensions, a stress propagation path is determined. The stress propagation path is an ordered sequence of at least one target dimension that starts from the stress source dimension, propagates along the ring topology, and passes through in sequence. Based on the stress propagation path control, a pre-set relaxation vernier slides on the annular topology, and at least one compressible target dimension is determined from the plurality of target dimensions according to the sliding position of the relaxation vernier; A step-by-step compression adjustment is performed on the resource configuration parameters corresponding to the compressible target dimension, wherein the single adjustment range of the step-by-step compression adjustment does not exceed a preset percentage of the current value.

2. The method for optimizing the allocation of test resources for a chassis testing machine according to claim 1, characterized in that, The multiple target dimensions include at least a fidelity dimension, a responsiveness dimension, a compliance dimension, and a consumption rate dimension; the operational status parameters of the fidelity dimension include the actual sampling rate, with the corresponding preset benchmark value being the ideal sampling rate; the operational status parameters of the responsiveness dimension include the actual latency, with the corresponding preset benchmark value being the target latency; The operational status parameters of the compliance dimension include disk placement success rate, with a preset baseline value of 100%; The operating status parameters of the consumption rate dimension include instantaneous power consumption, and the corresponding preset benchmark value is the power consumption threshold.

3. The method for optimizing the allocation of test resources for a chassis testing machine according to claim 1, characterized in that, The calculation process for the current stretch amount in each target dimension includes: For the fidelity dimension, the ratio of the difference between the ideal sampling rate and the actual sampling rate to the ideal sampling rate is calculated, and this ratio is truncated to the [0,1] interval to obtain the fidelity stretching amount; For the responsiveness dimension, the ratio of the difference between the actual delay and the target delay to the target delay is calculated, and this ratio is truncated to the [0,1] interval to obtain the responsiveness stretch; For the compliance dimension, calculate the difference between 1 and the success rate of the disk placement to obtain the compliance stretch amount; For the consumption rate dimension, the ratio of the difference between instantaneous power consumption and power consumption threshold to the difference between rated power consumption and power consumption threshold is calculated, and this ratio is truncated to the [0,1] interval to obtain the consumption rate stretching amount.

4. The method for optimizing the allocation of test resources for a chassis testing machine according to claim 1, characterized in that, The calculation process for the cumulative stress in each target dimension includes: For each target dimension, a fixed-length circular queue is maintained, the length of which corresponds to the preset duration divided by the sampling period; In each sampling period, the positive value obtained by subtracting the corresponding elastic limit from the current stretch amount is pushed into the circular queue, and the old value at the head of the queue is popped out. The sum of all values ​​in the circular queue is calculated as the cumulative stress for that target dimension.

5. The method for optimizing the configuration of test resources for a chassis testing machine according to claim 2, characterized in that, The coupling stiffness between the preset adjacent target dimensions is obtained through offline calibration. The order of the adjacent target dimensions on the ring topology is as follows: fidelity dimension, responsiveness dimension, compliance dimension, and consumption rate dimension, with the consumption rate dimension being adjacent to the fidelity dimension.

6. The method for optimizing the configuration of test resources for a chassis testing machine according to claim 5, characterized in that, The process of determining the stress propagation path includes: Calculate the first impedance value propagating clockwise from the stress source dimension and the second impedance value propagating counterclockwise, wherein the impedance value in either direction is obtained by dividing the coupling stiffness of the adjacent target dimension in that direction by the current tensile amount of that adjacent target dimension. The direction with the smaller impedance value is determined as the main propagation direction; Starting from the stress source dimension, adjacent target dimensions are checked sequentially along the main propagation direction. Target dimensions whose current tensile amount is greater than or equal to the corresponding elastic limit are added to the propagation path in sequence until the first target dimension whose current tensile amount is less than the corresponding elastic limit is encountered.

7. The method for optimizing the configuration of test resources for a chassis testing machine according to claim 6, characterized in that, Determining at least one compressible target dimension from the plurality of target dimensions includes: Maintain the current movement direction of the relaxed cursor and the reverse continuous count; In each sampling period, the main propagation direction corresponding to the current stress propagation path is obtained and compared with the current movement direction; If the main propagation direction is the same as the current movement direction, then the reverse continuous count is cleared to zero, and the relaxation cursor is controlled to slide one step along the current movement direction to move to the next adjacent connection point; If the main propagation direction is opposite to the current movement direction, the reverse continuous count is incremented by one. When the reverse continuous count reaches a preset number of times threshold, the current movement direction is reversed, the reverse continuous count is cleared to zero, and the relaxation cursor is controlled to slide one step along the reversed direction. Each time the relaxed cursor slides past a connection point, the target dimension located downstream of the main propagation direction among the two target dimensions connected by that connection point is marked as a compressible target dimension, and the compression window timing is started.

8. The method for optimizing the allocation of test resources for a chassis testing machine according to claim 1, characterized in that, The duration of the compression window is a preset fixed value. Within the compression window, the same compressible target dimension can be subjected to a maximum of one step compression adjustment.

9. The method for optimizing the allocation of test resources for a chassis testing machine according to claim 7, characterized in that, Performing step-by-step compression adjustments on the resource configuration parameters corresponding to the compressible target dimension includes at least one of the following operations: When the compressible target dimension is the fidelity dimension, the sampling precision is reduced first. After the sampling precision drops to the preset minimum precision, the sampling rate is reduced, with each reduction not exceeding 10% of the current value. When the compressible target dimension is a responsive dimension, the data pipeline is switched from a strictly ordered submission mode to an unordered submission mode with timestamp reordering. When the compressible target dimension is the compliance dimension, the data storage mode is switched from full storage mode to key frame plus differential storage mode. When the compressible target dimension is the consumption rate dimension, reduce the clock frequency of non-critical computing tasks, with each reduction not exceeding 10% of the current value.

10. The method for optimizing the allocation of test resources for a chassis testing machine according to claim 1, characterized in that, It also includes recovery steps: When the cumulative stress in all target dimensions is less than the preset stress threshold and the duration exceeds the preset stabilization time, the relaxation cursor is controlled to slide in the opposite direction of the current movement direction at a slower speed than that of the compression phase. Each time a connection point is traversed in reverse, the compressible marker of the downstream target dimension of that connection point is removed, and a step-by-step recovery adjustment is performed on that target dimension. The step-by-step recovery adjustment is the inverse operation of the step-by-step compression adjustment, and the recovery magnitude of a single operation does not exceed 10% of the current value. The stress state ends when the relaxed cursor returns to its initial position and the resource configuration parameters for all target dimensions are restored to their pre-compression state.

Images